A response-regulator-like activator of antibiotic synthesis ... - CiteSeerX

MiCrObiOlOgy (1998), 144,727-738

Printed in Great Britain

A response-regulator-like activator of antibiotic synthesis from Streptomyces coelicolor A3(2) with an amino-terminal

domain that lacks a phosphorylation pocket Ellen P. Guthrie,’t Christine S. Flaxman,2 Janet White,’ David A. Hodgson,2 Mervyn J. Bibb’ and Keith F. Chaterl

Author for correspondence: Keith F. Chater. Tel: +44 1603 452571. Fax: +44 1603 456844. e-mail: [email protected]

1

John lnnes Centre, Nonnrich Research Park, Colney, Nonnrich NR4 7UH, UK

* Department of Biological Sciences, University of Wawick, Coventry cV4 7AL, UK

In Streptomyces coelicolor A3(2), bldA mutants that lack the tRNA for the rare leucine codon UUA fail to make the red undecylprodigiosin antibiotic complex. To find out why, red-pigmented while bald (Pwb) derivatives of a bldA mutant were isolated. Using a cloning strategy that allowed for (and demonstrated) dominance of the mutations, they were localized to the red gene cluster. By using insert-mediated integration of a $C31 phage-based vector, one of the Pwb mutations was more precisely located between redstructural genes to a segment of approximately 1kb about 4 kb from the known pathwaylspecific regulatory gene redD. The segment contained most of an ORF (red2) encoding a protein (RedZ) with end-to-end similarity to response regulators of diverse function from a variety of bacteria. Remarkably, in Red2 hydrophobic residues replace nearly all of the charged residues that usually make up the phosphorylation pocket present in typical response regulators, including the aspartic acid residue that is normally phosphorylated by a cognate sensory protein kinase. A single TTA codon in red2 provided a potential explanation for the bldA-dependence of undecylprodigiosin synthesis. This codon was unchanged in three Pwb mutants, but further analysis of one of the mutants revealed a potential uppromoter mutation. It seems possible that a combination of low-level natural translation of the UUA codon by a charged non-cognate tRNA, coupled with increased transcription of red2 in the hnrb mutant, allows the accumulation of a threshold level of the RedD protein. Keywords : Streptomyces coelicolor A3(2), undecylprodigiosin, bldA, RedZ, RedD

INTRODUCTION Antibiotics and other secondary metabolites are among the most important products of bioindustry. It is therefore a matter of pressing interest to understand how production levels are regulated so that wellinformed approaches can be used for strain improvement and fermentation development (Chater, 1990 ; Baltz & Hosted, 1996). In streptomycetes, which produce many commercially important antibiotics, the regulation of secondary metabolism is also of fun-

...,.....,...........................................................,......................................................,,.......,........... tPresent addrass: New England Biolabs, Beverley, MA 01915-9990, USA. The EMBL accession number for the nucleotide sequence reported in this paper i s Y07902. 0002-1863 (0 1998 SGM

damental interest, because it is responsive to both developmental and environmental signals (Chater & Bibb, 1997). For studies of this subject Streptornyces coelicolor A3(2) provides an excellent model system because of its welldeveloped genetics and its ability to produce at least four quite different antibiotics. Among at least 21 loci known to have pleiotropic effects on production of most or all of these compounds (Champness & Chater, 1994; Bibb, 1996; Chater & Bibb, 1997; Chakraburtty et al., 1996; Marthez-Costa et al., 1996) are several bld genes whose action is also important for normal morphological differentiation (Champness & Chater, 1994).Most or all of these pleiotropically acting genes seem likely to influence the activity of pathway-specific regulatory genes (Bibb, 1996; Chater & Bibb, 1997). 727

E. P. GUTHRIE and OTHERS

In S. coelicolor A3(2) all antibiotic production (as well as normal aerial mycelium development) is prevented on most media by mutations in 6ldA (Merrick, 1976), which encodes the only tRNA that can efficiently translate the leucine codon UUA (Lawlor et al., 1987; Leskiw et al., 1991a). This codon is rarely used in Streptomyces spp., which contain more than 70 mol% GC in their DNA (Wright & Bibb, 1992). Indeed, the unimpaired growth and viability of 6ldA mutants shows that the TTA codon is not present in any essential gene and most of its known occurrences are in pathwayspecific genes, often regulatory, that are involved in antibiotic production. This has given rise to the idea that 6ldA exerts a regulatory role (Leskiw et al., 1991b), though searches for evidence of such a function have given differing results under different conditions (Leskiw et al., 1993; Gramajo et al., 1993). At least some of the genes encoding enzymes for biosynthesis of three of the S. coelicolor antibiotics are not transcribed in 6ldA mutants of S . coelicolor (Guthrie & Chater, 1990; Bruton et al., 1991;A. Wietzorrek & K. F. Chater, unpublished). In the case of actinorhodin biosynthesis the explanation is simple : actll-ORF4, encoding the pathway-specific activator of actinorhodin biosynthesis, contains a TTA codon and substitution of this codon by a different leucine codon resulted in 6ldAindependent production of blue-pigmented actinorhodin (Fernhdez-Moreno et al., 1991). The pathway for synthesis of another pigmented antibiotic of S . coelicolor, undecylprodigiosin (a tripyrrole compound produced as the major component of a mixture of red-pigmented prodigionines here termed Red; Tsao et al., 19SS), is positively controlled by the actll-ORF4 homologue redD (Narva & Feitelson, 1990; Takano et al., 1992), which is located at one end of the red gene cluster that encodes the enzymes for Red production (Malpartida et al., 1990; Hopwood et al., 1995). Although transcription of red pathway genes in most conditions appears to depend on both redD and 6ldA, the nature of this dual requirement is obscure because redD contains no TTA codons (Narva & Feitelson, 1990). Moreover, Red is produced at almost wild-type levels by 6ldA mutants on certain media, apparently in response to low phosphate levels (Guthrie & Chater, 1990;White & Bibb, 1997). Here we describe the use of a genetic approach to investigate further the reason for the 61dA dependence of red gene expression. The analysis of mutations that bypass this dependence has led to the identification of red& a gene in the red cluster that encodes an unusual homologue of a wellknown family of transcriptional activators. red2 contains a TTA codon and is therefore expected to depend on 6ldA for translation of its mRNA. METHODS Strains, media and microbiological methods. S. coelicolor A3(2) derivatives used were: M145 (prototrophic SCP1SCP2-) and M124 (proA1 argA1 cysD18 SCP1- SCP2-) (Hopwood et al., 1985);Jl50l (hisA1 uraA1 strA1 pgl SCP1SCP2-; Chater et al., 1982); 51700, a bldA39 derivative of

728

Jl50l (Piret & Chater, 1985); EG6 and EG9, red-pigmented (Pwb, 'pigmented while bald ') derivatives of 51700 isolated after UV irradiation (containing the Pwb-6 and Pwb-9 mutations, respectively); EG16, a Pwb-16 derivative isolated after treatment of 31700 with NTG; and EG311, E G l l l and EG441, Pwb-6-, Pwb-9- and Pwb-16-containing mthB2 strA1 bldA39 NF recombinants, respectively, from crosses of EG6, EG9 and EG16 with J668 (mthB2 cysD18 agaA7 bldA39 NF SCP2*) (Merrick, 1976).Propagation of 4C31 derivatives was on S. lividans 66. The rich agar medium R2YE and the minimal agar medium MM, and methods for subculturing strains, and making, transforming and regenerating protoplasts were as described in Hopwood et al. (1985);SMM was described by Takano et al., 1992; SY contained (w/v) 1.5% soluble starch, 0 1o/' Difco yeast extract, 0 1 OO/ K,HPO,, 0 1 YO MgSO, .7H20, 0 3 YO NaCl, 1-5YO Difco agar, pH 7.4. Standard conditions were used for the transformation and growth of Escherichia coli strains ED8767 (the host for routine subcloning using pBR vectors) and TG1 (used for subcloning using pUC and other lac2 a-containing vectors) (Sambrook et al., 1989). Plasmid and bacteriophage vectors, cloning procedures and DNA techniques. The Streptomyces vector for shotgun cloning of DNA from Pwb mutants was pIJ698 (Kieser 8c Melton, 1988), a low-copy-number SCP2*-derived vector containing the selectable tsr gene encoding thiostrepton resistance. PI5698 propagated in S. lividans was prepared for ligation by digestion with BglII and treatment with calf intestine alkaline phosphatase. In separate experiments it was then ligated with DNA from EG111, EG311 and EG441 after partial digestion of the donor DNA with Sau3AI to generate fragments of about 7-10 kb (fractionated on a sucrose gradient as described in Hopwood et al., 1985). The ligated DNA was used to transform protoplasts of S. coelicolor M124 to thiostrepton resistance as described in Hopwood et al. (1985). After sporulation (about 6 d) the colonies were replica-plated to R2YE plates spread with the bldA mutant 51700. After about 6 d these plates were replica-plated to M M containing histidine, uracil, streptomycin and thiostrepton to select for 51700 transconjugants containing PI5698 (potentially with DNA inserts) and to counter-select the M124-based donor. After a further 6 d these were replicated to MM with histidine or uracil or both supplements to distinguish between pigmented colonies able to grow on singly supplemented plates that must have arisen from chromosomal recombination between M124 and 51700, and those probably arising from plasmid transfer (retaining both auxotrophic requirements). Standard procedures (Hopwood et al., 1985) were used to isolate and analyse plasmid DNA from selected strains. The presence of HindIII sites flanking the BglII cloning sites of PI5698 and the rarity of such sites in Streptomyces DNA made it straightforward to subclone the whole of each insert into the E. coli vector pBR327 (Covarrubias & Bolivar, 1982) for physical analysis and further subcloning. Where DNA was to be re-introduced into S. coelicolor derivatives, it was subcloned either into pIJ698 or 4C31 KC861 (Bruton et al., 1991) with S. lividans as an intermediate host to bypass the restriction barrier observed when DNA from a methylating E . coli host (Dam+or Dcm+or EcoK+)is introduced directly into S. coelicolor A3(2) (Kieser & Hopwood, 1991).The wild-type red DNA-containing plasmids pCLLl (vector pBR322, insert 16.4 kb EcoRI fragment including red& redF and redE; Feitelson et al., 1985; Malpartida et al., 1990; DNA supplied by J. Feitelson) and pIJ2341 (vector pBR329, insert 13-7kb PstI fragment including redD ; Malpartida et al., 1990) were used as probes in Southern blotting (Hopwood et al., 1985).

Regulation of undecylprodigiosin production The low-copy-number plasmid PI~2351,containing the right ca 50% of the red cluster in the vector pIJ922 (Malpartida et al., 1990), was used to evaluate the effects of extra copies of redD and red2 on Red production by 51700. DNA sequencing. Single-stranded DNA templates were sequenced by the chain-termination technique (Sanger et al., 1977) using defined restriction fragments cloned in M13 mp18/19 (Yanisch-Perron et al., 1985) and in some cases specially designed oligonucleotideprimers. The sequence was determined on both strands and analysed using the program FRAME (Bibb et al., 1984). Database searches used the BLAST (Altschul et af., 1990) and FASTA (Devereux et al., 1984) programs. 51 nuclease protection assays. A 1-2kb BamHI-EcoRI fragment (sites 1-4 in Fig. 3 of White & Bibb, 1997),uniquely labelled at the 5' end of the EcoRI site located in the red2 coding region using [y-32P]ATPand T4 polynucleotide kinase (Sambrook et al., 1989), was used to map the redZ transcriptional start site. About 002 pmol (approx. lo4 Cerenkov counts min-l) of labelled probe was hybridized to 60 pg RNA in sodium TCA buffer (Murray, 1986)at 45 "C overnight after denaturation at 65 "Cfor 15 min. All subsequent steps were as described by Janssen et al. (1989). The RNA-protected fragment was subjected to electrophoresis alongside a sequence ladder generated using a 19 nt primer whose 5' end corresponded to the 5' end of the labelled EcoRI site used for S1 nuclease mapping. Phylogenetic analysis. The program CLUSTAL w (Thompson et al., 1994) was used to align the sequences. Distances were calculated using PROTDIST, which employs maximum likelihood estimates based on the Dayhoff PAM matrix. The tree was constructed using FITCH, which employs the FitchMargoliash method, and displayed with TREEVIEW (Page, 1996). PROTDIST and FITCH were from the PHYLIP package, version 3.5 (Felsenstein, 1989). RESULTS Identificationof mutants showing the Pwb (pigmented while bald) phenotype

To investigate the role of bldA in controlling antibiotic production, we sought mutants of the normally unpigmented bldA39 mutant J1700 that produced blue (yactinorhodin; Bystrykh et al., 1996) or red (undecylprodigiosin) antibiotic pigments on R2YE medium. Unexpectedly, deep red colonies were very frequent. These nearly all turned out also to be sensitive to chloramphenicol (Cm"). A high frequency of CmS mutants with deep red mycelium has been noted previously in morphologically wild-type S. coelicolor (Sermonti et al., 1977) and is associated with DNA rearrangements (Altenbuchner & Cullum, 1984;Dyson & Schrempf, 1987). Although an explanation of the red pigmentation of such mutants would be of interest, their complex nature led us to avoid them in this analysis. We therefore screened again for mutants showing the Pwb phenotype, this time incorporating chloramphenicol (10 pg ml-l) into the medium. This caused a dramatic drop in the frequency of Pwb mutants, necessitating the use of mutagenesis. Even after mutagenesis we could find no blue (y-actinorhodin-producing) colonies, but several showing moderate red coloration were obtained

among the ca 1800 colonies screened after UV treatment and ca 2100 colonies after NTG treatment. We selected EG6 (Pwb-6) and EG9 (Pwb-9) (isolated following UV irradiation) and EG16 (Pwb-16) (isolated after NTG treatment) for further study (Fig. la). (In this paper we found it convenient to use the Pwb designation equally to describe phenotype, mutants and mutations ;we have avoided its use in the italicized form, pwb, as a genetic designation, preferring the term redZ for the gene eventually identified through the Pwb mutants.) The mutants all showed perceptible red pigmentation on SY medium but EG6 differed from EG9 and EG16 in failing to develop pigment on M M with either mannitol or glucose as carbon source. Attempts to establish genetic map locations for the three mutations in crosses with a differently marked bldA mutant (5668; Merrick, 1976) did not give clear-cut results : although the phenotypic difference between the Pwb mutants and the 51700 parent was reproducible, intermediate or low levels of pigmentation were obtained among some recombinants, which could therefore not be scored satisfactorily. In the course of these crosses, some recombinants showing an unambiguous Pwb phenotype were retained, including the three strains EG311 (Pwb-6), E G l l l (Pwb-9) and EG441 (Pwb-16) used as sources of DNA for cloning. Cloning of Pwb DNA suggests a location for hnrb mutations in the red cluster Cloning of the Pwb mutations was expected to provide access to DNA involved in red gene regulation. Because we did not know whether the Pwb mutations would be dominant or recessive, we used a cloning strategy that allowed the comparatively easy recognition of the desired clones in either case. The approach was to make a library of DNA from each Pwb mutant in the lowcopy-number transmissible Streptomyces plasmid vector pIJ698 and then to transfer the library from its primary host (M124) into the unpigmented bZdA mutant 51700 using plate mating. The transconjugant colonies were scored for the Pwb phenotype on the presumption that they would be uniformly red if the Pwb mutations were dominant and specked with red if the mutations were recessive (because of the occasional production of Pwb/Pwb homogenotes by gene conversion). In fact, a few uniformly red transconjugant patches were obtained for each mutant (Pwb-6,4/4800 colonies ;Pwb-9,8/1760 colonies ;Pwb-16,2/1200 colonies), suggesting that the Pwb mutations were dominant. For each mutant, plasmid DNA was isolated from one of the pigmented J1700 transconjugant colonies and designated PIJ Z 3 0 (Pwb-6), pIJ2520 (Pwb-9) and PIJ254O (Pwb-16). When each DNA sample was used to transform 51700, nearly all transformants were pigmented, verifying that the Pwb phenotype was associated with the plasmids and reinforcing the suggestion that the mutations were dominant. The inserts from the three apparently Pwb-containing plasmids were subcloned as Hind111 fragments into pBR327 in E. coli, yielding pIJ2.531 (Pwb-6), pIJ2.521 (Pwb-9) and pIJ2.541 (Pwb-16), and analysed by BamHI 729

E. P. G U T H R I E a n d O T H E R S

Fig. 1. (a) Red production by EG9 (bldA39, Pwb-9) and its absence from the parental strain J1700. The bMA+ strain J1501 (the parent of J1700) produces both Red and the bluepigmented actinorhodin and sporulates. R, right-hand panel photographically reversed to facilitate comparison. The cultures were grown on R2YE for 5d. (b) Strong Red production resulting from introduction of cloned Pwb DNA into J1700 ( M A ) ; much weaker pigmentation is induced by plJ2351 containing a segment of wild-type red DNA that includes redCDZ as well as the regions present in all of the Pwbcontaining plasmids. The strains were grown for 5 d on MM supplemented with histidine, uracil, streptomycin and thiostrepton.

digestion. Several fragments appeared to be common to all three inserts and were similar in mobility to fragments present near the right-hand end of the red cluster. Proof that the plasmid clones indeed contained red DNA came from Southern blotting using the red-containing plasmid pIJ2341 (Malpartida et al., 1990) as probe. All three clones contained a common region of the red cluster (Fig. 2) that included the pathway-specific regulatory gene redD (Narva & Feitelson, 1990; Malpartida et al., 1990). It was possible that the Pwb phenotype caused by the clones might have been attributable to an additional copy of redD (or other red genes) rather than to cloned Pwb DNA. We therefore introduced pIJ23.51, a pIJ922 derivative containing a segment of wild-type red DNA 730

that spanned the inserts present in the three clones, into 51700 (pIJ2351 is based on the low-copy-number SCP2" replicon also present in the pIJ698 vector used for Pwb cloning). Slight pigmentation resulted, but it was much weaker than was seen with the Pwb clones (Fig. lb), suggesting that the cloned DNA did indeed contain the Pwb mutations. Use of insert-directed integration of dC31 vectors to localize the Pwb-9 mutation

Partial Sau3AI digests (fragment sizes ca 4 kb) of the Pwb-9-containing insert of PIJ2520 were used to provide quasi-random fragments for insertion into the BamHI

Regulation of undecylprodigiosin production plJ2540 plJ2530 plJ2520

.............,....................................................................................................................................................................................................................................................,.............................................

Fig. 2. Restriction map of the part of the red cluster containing the Pwb mutations. The lower part shows details of subcloning of Pwb-9-containing DNA into &31 KC861 after Sau3Al partial digestion of the plJ2520 insert. Solid lines indicate the approximate lengths of the inserts, vertical lines indicate restriction sites known to be present in each insert and dotted lines indicate the extent of uncertainty about the precise positions of the insert ends. The effects of the inserts on Red production in J1700, J1501 and EG9 are shown to the left of the map. Inserts giving a Red+ phenotype in 11700 presumably carry the Pwb-9 mutation, while those giving a Red- phenotype on both J1501 and EG9 are likely t o reflect disruption of a red transcription unit. The chromosomal segments present in plJ2520, plJ2530 and ~112540,and the extents and directions of the coding regions of redZ (this paper) and redD (Narva & Feitelson, 1990) are shown above the map. The chromosomal insert in plJ2351 includes the entire mapped region and additional flanking sequences (Malpartida et a/,, 1990). The shading indicates the region that is common t o all of the inserts that confer the Pwb phenotype on J1700 (with the exception of isolate 19) and that presumably contains the Pwb-9 mutation. The extent of the nucleotide sequence shown in Fig. 3 is indicated by the triangles and the BamHl fragment previously used t o define redX (Guthrie & Chater, 1990) is marked by circles. B, BamHI; €, EcoRI; G, Bglll; N, Ncol; PI hull; S, Sphl; X, Xhol.

site of KC861, a Streptomyces vector based on an attPsite-deleted derivative of the temperate phage 4C31 (Bruton et af., 1991). Phages containing inserts were able to lysogenize S. coelicolor strains by single crossover integration at the homologous chromosomal region. If the insert contained a functional, dominant Pwb mutation, then at least some lysogens of the 6ldA mutant 51700 should have shown the Pwb phenotype (unless the insert also disrupted a red transcription unit needed for pigmentation).

a common segment of approximately 1 kb located about 4 kb to the left of redD. [This segment was also present in one of the phages (isolate 19) that did not give rise to the Pwb phenotype. One explanation for this disparity could be recombination of the insert with the homologous region of the host chromosome to restore the wild-type sequence, which might have taken place during replication of the 4C31 derivative in S. Zividans (which contains a red cluster; Malpartida et al., 1990) during phage propagation.]

Of 26 insert-containing phages tested, 11 caused pigmentation, further indicating that the cloned Pwb DNA was dominant. DNA isolated from a representative of each of the 26 types of lysogen was analysed by Southern blotting after EcoRI, BamHI, XhoI, NcoI and SphI digestion with PI 52341 as probe. DNA of the 26 phages was also subjected to restriction analysis with EcoRI, BamHI and PvuII. The analysis was somewhat complicated by the use of Sau3AI as the enzyme for subcloning into KC861, but clear-cut results were obtained for 16 of the 26 phages (Fig. 2). The data confirmed the expected integration of the prophages into the red region and gave information on the extent, localization and orientation of the inserts. The inserts giving rise to the Pwb phenotype overlapped, containing

T o test whether any of the 26 insert-containing phages might contain fragments internal to red transcription units, they were also introduced into Jl501, the bfd' (hence Red-producing) parent of 51700, and into the Pwb mutant EG9 (Fig. 2). Of the 16 well-defined phages, isolates 9,12 and 25 caused a loss of red pigmentation in 51501 and EG9, suggesting that they, and no other phages, contained inserts internal to red transcription units. Of these, numbers 9 and 12 contained inserts from the left end of the region, and number 25 an insert from the region between the ' common Pwb' region and redD. The insert in number 25 overlaps with a BamHI fragment shown previously - on the same criterion - to be internal to a biosynthetic transcription unit termed redX (Guthrie & Chater, 1990). Thus, the Pwb-9

731

E. P. G U T H R I E and O T H E R S

u7 d.

75

f

150 5 225

30

300

55 375 80 450 105 525

130 600 155 675

180 750 825 205 217 900 975 1050

Figrn 3. Nucleotide sequence of the redZ region. The sequence is oriented to correspond to Fig. 2 and extends between the marked Bglll and EcoRl sites shown in that figure. The deduced amino acid sequence of RedZ is shown in single letter code above the nucleotide sequence. The position in the EcoRl site used for labelling the probe used in 51 mapping is shown by an arrowhead and the transcription start point is marked by a solid circle. The base change in the cloned Pwb-6 mutant DNA is shown at nucleotide position 62. Probable -10 and -35 hexamers for recognition by RNA polymerase holoenryme are underlined, as is the TTA (leucine) codon at positions 675477. Amino acids that replace the highly conserved residues of response regulators that appear to play an essential role in phosphorylation are boxed. The putative helix-turn-helix motif of RedZ is overlined. An inverted repeat that may play a role in termination of both the redZ and redY transcripts is shown by converging arrows above the nucleotide sequence (positions 923-945).

mutation appears to be located between two red biosynthetic transcription units. A gene in the hnrb region contains a l l A codon, potentially a target for the action of bldA

The region common to all the Pwb-%containing phages consisted largely of an EcoRI fragment of approximately 800 bp. Accordingly, this fragment was subcloned from wild-type DNA present in pIJ2341 into M13 mp18 and sequenced. The sequence contained two converging protein-coding sequences, both truncated at their 5' ends. Further sequencing of the adjacent BglII-EcoRI fragment that contained most of the sequence common to the Pwb-containing phages revealed the 5' end of the longer of the two incomplete coding sequences found initially. The whole sequence is shown in Fig. 3. The central part of the coding sequence deviates somewhat from typical protein-coding DNA of streptomycetes. Usually, such DNA contains about 72 mol% GC and the third base in at least 90% of codons is a G or C, whereas the central ca 40 Yo of this gene contains about 60 mol YO GC and only 70 YO of codons end with G or C, with several very rare codons being used. Most notably, 732

redZp

-

T C G A

/i

\:

T*

G

Fig. 4. Determination of the transcription start site of redZ. T, C, G and A, redZ nucleotide sequence ladder; M145, RNA-

protected fragment derived from RNA isolated from a culture of M145 grown to late stationary phase in SMM. The asterisk indicates the most probable transcription start site; the sequence given is that of the template strand.

the 'low GC' region ends at a TTA codon, providing a potential target for the action of bldA and therefore a possible explanation for the Red- phenotype of bldA mutants. This, and the absence of any other complete ORF from the Pwb region, suggested that the gene, termed red2, was responsible for the Pwb phenotype. S1 nuclease mapping was used to detect the 5' end of red2 mRNA isolated from a liquid culture of S. coelicolor strain M145 grown to early stationary phase in SMM. The DNA probe was uniquely labelled at residue 263 of the strand complementary to that shown in Fig. 3 using an EcoRI-generated end. A single protected fragment of i66 nt was generated (Fig. 4), corresponding to an apparent transcription start point at nucleotide 98. Thus redZ mRNA has an untranslated leader sequence of 113 nt. The transcription start point is preceded by appropriately located sequences resembling the consensus - 10 and -35 regions for promoters presumed to be recognized by the major Streptomyces RNA polymerase holoenzyme (Brown et al., 1992; Strohl, 1992). Further sequence analysis (N. Hartley & M. J. Bibb, unpublished) confirmed the presence of another protein-coding region (redV) terminating 153 bp upstream of the redZ transcription start site. The redZ gene product is homologous to a family of regulatory proteins

The deduced RedZ amino acid sequence was compared with known proteins using the FASTA (Devereux et al., 1984) and BLAST programs (Altschul et al., 1990). RedZ

-

Regulation of undecylprodigiosin production a2

Dnrn-Strpe R m - S trco Bvga-Borpe Trpo-Pseae Uhpa-Ecoli Uvry-Ecol i Dew-Bacsu Gere-Bacsu AbsaZ-Strco Orfl-Strli Yfik-Bacsu Yxjl-Bacsu Narp-Ecoli Redz-Strco

. . .

.MTI . .MGEMV . . . .MYN

.... ....

.MS

.MIKT:I

....

E

C

L TG P E A G

Wry-Ecoli

Degu-Bacsu Gere-Bacsu

Yfik-Bacsu Yxjl-Bacsu Narp-EcoL i Redz-Strco

a

P Z FT

DSID'K

-

+

AGEMRTTNPD TRRILTGAPD TKLITSSLPS AKDIMSALPN LNALRRD-GV L T E L . . . .AR

. . _ . .

T A V T T

A. PR .K .K AQ

V V V I I

V I L V I

I

L C K T L

Bvga-Borpe Trpo-Pseae Uhpa-Ecoli Uvry-Ecol i Degu-Bacsu Gere-Bacsu Absa2-Strco Orfl-Strli

Yfik-Bacsu Yxjl-Bacsu Narp-Ecoli Redz-Strco

RV R F TY IH C Y RY SY

. .

. . . .

FDTELTLAAW LDETLTVALL FPSTTLSEMR FPTGALRSIN LTPDIAIKLA IASDIAQQMA LHPKVTHNLV

..........

sc LSPLVARRLM

AL MV I R A A A A1 IRAGA KV I R T%?T E L NA 1

LAPAITRRLV LPPELTAQML FRTVTAAKII FSERVNQYLR IAPEEAQEAL

60

.

.

+

.

.........

a5

1

GEDRYIERA DLDDLVYAA KDSALIAEA DTEEYVFEGL DASSDVFALT ENTHRAFERB

'

. . . . . . . . . .

a

a5 Dnrn-S trpe R m - S trco

0 55 57 55 59

ELARK EVVLR EKAKE D I I ALAKQ SRIDLANR . S LATAAE

a4

P4

I C . R T L R F A R'L A T V R Q S a AGRLRARYGD .RLV T A K R P G V L R R A F' IARL-QSLGL PLRVL 'GQ P P S L F A R R C 1ARL.KSLKL DTKVL RQ NRSQFARR. LSQLPKGMA. . . .TI VH DSPALVEQA TRKIARSTAD V .KII V H T E N P L P A KV:M! T K Q L V E L Y P E S .K V I IH DDENYVTHAH

..........

Absa2-Strco Orfl-Strli

57

.....

I

a3 Dnm-Strpe Ramr-S trco Bvga-Borpe Trpo-Pseae Uhpa-Ecoli

56 57 55 54 55 55

.MI

. . . . .MI . . .MTKV . . . . . . . . . .

....

I

I

HSELIRTVLE DADGPELSAA IDGLKIARE

GNTGCP . . . . . . . . . . . . . . KGAEMP . . . . .......... M G D N A . . . . . . . . . . . . . . . -QQSR..... . . . . . . . . . . SGRQDP . . . . .......... LSQIEP.. . . . . . . . . . . . . NEFRRLATSG VSAHPQHEVY . . . . . . . . . . . . . . . . . .M K TELRRAP..S PRSEV . . . . . ERHAHTT..P PAPNT . . . . . NEWKREKQLK GINEI . . . . . SETFRAKQQT HAEEL . . . . . ER . . . . . . . . . . . . . . . EMF TRYWRSK . . . . . . . . . . . . .

SSPPA NAPAH HENLHE V I GKPLR R . RCSPD GAAPQ EMDAD

.....

ASDPR DVTPE DMSAD DTLPE DSDPE ETSVE

. . . . . . . . .

115 115 114 111 111 114 116

. .

0 114 117

SG

114

118 119 110

~

HV

QL TL QL QY EK LM QM EL

QL AE QQ HE LL

152 152 159 155 148 160 176 22 165 169 168 167 164 156

. . . . . . . . . .

202

.........

KSDSTLISV DNEARMLES

.........

EKT ES P F A S PEIRRPLHI EKEFQSKPS . S G E R T .T L .TAHRDLQT .EKPNELLD .AEP.....F GAEEDPFSV .A P K N L R P

GM

EB

VRIDSMS

54

A

. . . . . . . . . .

Trpo-Pseae

. . . . . . . .

I

I

A. Q '

TILPLQQFGA VAIDYES a LL

. . . . . . 220 . . . . . . . 218

Q

G

.........

215 I C P G Y G T PAR217

H-T-H

....................................................................................................................................,......................................................................................... ........... ................................................................. I..

.I..

Fig. 5. Alignment of RedZ with the most closely related response regulators found in the databases. The positions of the highly conserved amino acids that appear t o play an essential role in phosphorylation in this family of proteins, but that are absent from RedZ, are marked by triangles. The secondary structural features of the corresponding regions of CheY (Volz, 1995) discussed in the text are shown above the N-terminal domain. The putative helix-turn-helix motif (residues 163-182) is underlined. The sequences were aligned and displayed using the PILEUP and P R E ~ B O programs, X respectively, of the UWGCG package (Devereux eta/., 1984). The origin of amino acid sequences and their database accession numbers (SWISS-PROT unless stated otherwise) are: Dnrn-Strpe, S. peucetius (EMBL L37338); Ramr-Strco, S. coelicolor (GenBank U03771); Bvga-Borpe, Bordetella pertussis (P16574); Trpo-Pseae, Pseudornonas aeruginosa (P24908); Uhpa-Ecoli, E. coli (P10940); Uvry-Ecoli, E. coli (P07027); Degu-Bacsu, Bacillus subtilis (P13800); Gere-Bacsu, B. subtilis (P11470); Absa2-Strco, 5. coelicolor (EMBL US1332); Orfl-Strli, S. lividans (GenBank U63847); YfikBacsu, B. subtilis (EMBL D78508); Yxjl-Bacsu, B. subtilis (P55184); Narp-Ecoli, E. coli (P31802); Redz-Strco, S. coelicolor (EMBL Y07902). Black boxes indicate positions in the alignment where the same amino acid is found in a t least 8 of the 14 sequences; grey boxes indicate amino acids that are similar to those marked in black.

is related to a large family of regulatory proteins, many of which are response regulators partnered by sensory histidine protein kinases in so-called two-component systems (Fig. 5; Hackenbeck & Stock, 1996). The

similarities extend through both of the characteristic domains of response regulators : the N-terminal domain that typically interacts with the sensor kinase, leading to phosphorylation of the response regulator; and the C-

733


Fig, 6. Phylogenetic tree of the presumptive DNA-binding domains found in class 3 response regulators (Pa0 & Saier, 1995), including RedZ and other streptomycete homologues. The amino acid sequences analysed correspond to positions 146-205 of RedZ shown in Fig. 5. The DNA-binding domain of AfsQ1 (AfsQ1-Strco, SWISS-PROT Q04942), a response regulator from S. coelicolor that belongs t o class 2, not class 3, response regulators is given as an outgroup. Branch length is proportional t o phylogenetic distance. The scale bar represents the number of substitutions per site during the evolutionary period represented by that length of branch. The origin of amino acid sequences not shown in Fig. 5 and their database accession numbers are: MalT-Escco, E. coli (P06993); AlkS-Pseol, Pseudomonas oleovorans (P17051); AgmR Pseae, P. aeruginosa (P29369); LuxR-Vibfi, Vibrio fischerii (P12746); SdiAEscco, E. coli (P07026); NarLEscco, E. coli (P10957); ComABacsu, B. subtilis (P14204); BrpAStrhy, 5. hygroscopicus (QOllO8); FimLEscco, E. coli (P21502); RcsBEscco, E. coli (P14374); RcsAKleae, Klebsiella aerogenes (P05338); NodW-Braja, Bradyrhizobium japonicum (P15940); FixJRhime, Rhizobium (Sinorhizobium)meliloti (P10958); FixJ-Rhime, Br. japonicum (P23221).

terminal domain that contains an a-helix-turn-a-helix motif responsible for binding to specific DNA sequences in or near promoters, leading to gene activation or repression, In its N-terminal domain, RedZ shows moderate similarity to the cluster 3 sub-group of receiver modules of response regulators (Pao & Saier, 1995), the most similar being UhpA (27YO identity), NarP (27 YO) and UvrY (28%) of E. coli, TrpO (25%) of Pseudomonas aeruginosa and BvgA (28%) of Bordetella pertussis (all percentage identities are for a region of overlap of 109 residues; Fig. 5). Cluster 3 receiver modules are generally, but not always, associated with DNA-binding domains classified as family 3 by Pao & Saier (1995). RedZ conforms to this tendency; its Cterminal putative DNA-binding region shows considerable similarity to family 3 DNA-binding domains, including DnrN (467 0 )AbsA2 ~ (42YO)and RamR (42YO) from streptomycetes, DegU (44%) and GerE (45%) from Bacillus subtilis and MalT (37YO) from E . coli (all percentage identities are for regions of overlap of 60 residues). In a detailed phylogenetic analysis (Fig. 6; Thompson et al., 1994; Felsenstein, 1989) the putative DNA-binding domains of RedZ, DnrN, AbsA2 and RamR, all from streptomycetes, formed a discrete subcluster. At least one family 3 response regulator from a Streptomyces sp. (BrpA) was well separated from this 734

subcluster, showing that the clustering is not a trivial consequence of the high GC content of Streptomyces genes (Fig. 6). The receiver-like domain of RedZ did not appear to be particularly closely related, in evolutionary terms, to any other cluster 3 receiver module (data not shown). Surprisingly, RedZ lacks several of the highly conserved residues that appear to be essential for phosphorylation of the receiver modules of response regulators. The invariant aspartate at position ca 55, which is the residue phosphorylated by the cognate histidine protein kinases, is replaced by valine; and the highly conserved lysine at position ca 100, which is generally required for efficient phosphotransfer, and which would usually be hydrogenbonded to the unphosphorylated aspartate at position ca 55, is replaced by an alanine. Moreover, one of the pair of aspartates conserved at positions ca 11 and 12 is absent from RedZ. These two aspartates would usually be embedded in close apposition to the conserved aspartate/lysine pair, playing important roles in phosphorylation. Thus, the structure of this part of RedZ may be maintained by hydrophobic interactions, in contrast to the hydrogen bonding and Mg2+ contacts that determine the structure of the phosphorylation pocket in typical response regulators (Volz, 1995).

Regulation of undecylprodigiosin production A Pwb mutant has a base change in the -35 region of the redZ promoter

Using PIJ2520, PIJ2530 and PIJ2540 as sources of DNA, the Pwb mutant copies of the 796 bp EcoRI fragment containing most of red2 were sequenced. No base changes were detected and the TTA codon present in the wild-type sequence was retained in each of the DNAs derived from the mutants. Analysis of the sequence upstream of this from Pwb-6 DNA revealed a base change in the -35 region of the redZ promoter (sequences upstream of the coding regions of redZ in Pwb-9 and Pwb-16 DNA have not yet been determined). This change caused a closer resemblance to the likely consensus -35 sequence for promoters presumed to be recognized by the major vegetative form of RNA polymerase holoenzyme : TCTCCA Pwb+ TTTCCA Pwb-6 TTGACA

-35 consensus

T o explain how this change might give rise to the Pwb phenotype, we postulate that it may improve interaction of the red2 promoter with RNA polymerase to give higher levels of the red2 transcript. Assuming a low degree of natural mis-reading of the UUA codon by charged non-cognate tRNAs, increased red2 mRNA levels would lead to proportionately greater production of RedZ protein, presumably to levels sufficient to activate RedZ target genes. Indeed, the presence of multiple (ca 10 copies) of red2 in a 62dA mutant restores Red production, consistent with the notion of a low level of natural mis-translation (White & Bibb, 1997). Similar observations were also made with the TTAcontaining gene actll-ORF4, which regulates actinorhodin production (Passantino et al., 1991; FernandezMoreno et a!., 1991). Quantitative analysis of red2 transcription in all three Pwb mutants is currently in progress. DISCUSSION The nature of the RedZ protein

Through the isolation of suppressor mutations causing pigment production in a b2dA mutant, we discovered re&, a gene encoding a response regulator-like protein, within the red biosynthetic cluster. Where sufficiently studied, all known end-to-end homologues of RedZ are subject to phosphorylation by a cognate histidine protein kinase, an interaction involving an N-terminal ‘receiver module’ domain of the protein. A few proteins consisting only of detached receiver domains are also known and the detailed structure of one of these, CheY, has provided the basis for much of the current modelling of receiver domains in more complex proteins. Here we use the CheY co-ordinates to compare residues conserved in the receiver domains of response regulators with those in equivalent positions in RedZ. Volz (1995) presented a table of the structural roles of conserved residues of receiver modules. Most of these

residues are of similar character in RedZ (Fig. 5), such that it should retain the overall @/a)S structure of receiver domains. The most striking changes are those affecting the part of the protein that, in response regulators, is involved in phosphorylation ; the normal hydrogen-bonding and Mg2+binding that characterizes this pocket (Volz, 1995) would be replaced by hydrophobic interactions, preventing all possibility of conventional phosphorylation. However, the N-terminal domain of RedZ should retain the ability to take part in the general kinds of protein-protein interactions common to all receiver modules attached to DNA-binding domains. Typically these would include association with sensor kinases, intramolecular interaction with the DNA-binding domain, perhaps homo- (or hetero-) oligomerization and interaction with the transcription initiation complex at target promoters. We speculate that RedZ exhibits an unusual variation on the theme of receiver modules in which the normal phosphorylating pocket, made up of elements from different parts of the primary sequence, has become entirely reconstructed to fulfil a different biochemical function. For example, it could provide a binding site for a small co-regulator molecule that contains a hydrophobic feature, as in the case of LuxR which has an unrelated, N-acyl homoserine lactone-binding N-terminal domain coupled to its RedZ-like C-terminal DNA-binding domain (Pao & Saier, 1995). Alternative hypotheses, including the possibility that the N-terminal domain has no regulatory influence on the activity of RedZ, cannot be excluded at this stage. The C-terminal domain of RedZ shows up to 46% identity with the sub-group of DNA-binding domains of response regulators known as family 3 (Pao & Saier, 1995). Within this sub-group, RedZ and several other streptomycete proteins involved in antibiotic production or morphological differentiation cluster together by themselves (Fig. 6 ) . This clustering suggests a comparatively recent origin of the genes for these response regulators from a common ancestral gene that may have controlled a much simpler range of stationary phase options. The significance of RedZ in the regulation of undecylprodigiosinbiosynthesis

The regulatory roles played by RedZ-like proteins, the presence of a TTA codon in the redZ coding region and the discovery of a base change in the red2 promoter of one of the mutants all provide strong circumstantial evidence that red2 is the target for 62dA in the red cluster. We have shown elsewhere that red2 is directly or indirectly required for transcription of red biosynthetic genes (White & Bibb, 1997). Its transcripts were present during exponential growth and increased in amount during transition and stationary phase, while transcription of redD, the previously identified pathwayspecific regulatory gene, was confined to the two latter stages of growth. Moreover, transcription of redD was highly dependent on red2, whereas mutation of redD had no effect on redZ transcription, suggesting that 735


gene has been discovered. This gene ( a d - O R F 4 ) encodes a RedD-like protein. The additional component in the red gene cluster, RedZ, seems to offer opportunities for further levels of regulatory signal input (Fig. 7) at the levels of r e d 2 transcription [is it repressed by AbsA2 (Brian et al., 1996) and is its transcription affected by other pleiotropic regulatory genes ?I, translation (which is predicted to be 6ldA-dependent) and RedZ activity (which might be modulated post-translationally).

Fig. 7. The involvement of redZ in Red biosynthesis provides multiple opportunities for the input of diverse signals. Bold arrows trace the pathway-specific regulatory system leading from red2 via redD to the expression of red structural genes (White & Bibb, 1997; Narva & Feitelson, 1990). Dashed arrows from shaded ovals indicate the major possible sources and targets of signal input that might influence the system. The putative co-regulator is proposed t o interact with the hydrophobic pocket of RedZ that replaces the phosphorylation pocket of typical response regulators. The pleiotropic regulators referred to include various abs, aba, afs, bld and other genes (see Introduction). Among the signals to which they might respond are ppGpp (Chakraburtty e t a/., 1996; Chakraburtty & Bibb, 1997), extracellular autoregulators (E. Takano, personal communication) and phosphate levels, although the restoration of Red production under conditions of phosphate-limitedgrowth (Guthrie & Chater, 1990) may reflect e tevated levels of mistranslation rather than a specific regulatory role (White & Bibb, 1997). We also speculate that the activity of RedD may itself be susceptible to modulation by some such signal input.

RedZ is a transcriptional activator of redD. Transcription of redZ appeared to be unaffected in a 6ldA mutant, but redD mRNA was undetectable, consistent with translational dependence of r e d 2 on bZdA and transcriptional dependence of re& on re&. The finding that Red synthesis involves more than one positively acting pathway-specific regulatory gene contrasts with actinorhodin biosynthesis in the same organism, where only one pathway-specific regulatory 736

It is possible that the presence of more than one regulatory gene in the red cluster may reflect the complexity of the converging biochemical pathway for Red biosynthesis (COCOet al., 1991). Such complex pathways may involve a modular genetic origin with different parts of the pathway having been brought into a common genetic background by lateral transfer of gene sets that had evolved separately. If so, each part of the pathway is likely to have co-evolved with its own pathway-specific regulatory system. Although transcription of redD depends, directly or indirectly, on redZ (White & Bibb, 1997), it will be interesting to determine whether RedZ and RedD independently activate different red genes or, at the other extreme, whether they are simply steps in a linear regulatory cascade. The latter seems to be the case in the closest parallel to the redZ/redD situation : the S . peucetius genes for daunorubicin production are transcriptionally dependent on DnrI, a RedD homologue and the dnrZ gene is itself transcriptionally dependent on dnrN, which encodes a response regulator homologous to RedZ (StutzrnanEngwall et aZ., 1992; Madduri & Hutchinson, 1995; Otten et al., 1995;Figs 5 and 6 ) .Although DnrN retains the aspartyl group which would be phosphorylated by interaction with a cognate sensor kinase, as well as most of the other conserved features of the active site, there is no nearby gene specifying an obvious potential protein kinase and a mutation of the potentially phosphorylated aspartyl residue had only a small effect on daunorubicin production (Otten et al., 1995). Thus, DnrN and RedZ may both fulfil their activating roles in an unphosphorylated state.

ACKNOWLEDGEMENTS We thank Mark Buttner and David Hopwood for comments on the manuscript. This work was supported by grants from the Agricultural and Food Research Council and the Biotechnology and Biological Sciences Research Council to the John Innes Centre; from the John Innes Foundation to E. P. G. ; from the EU BIOTECH programme (contract B102-CT942067) and by a BBSRC studentship to C. S. F.

REFERENCES Altenbuchner, 1. & Cullum, J. (1984). DNA amplification and an

unstable arginine gene in Streptomyces lividans 66. Mol Gen Genet 195, 134-138.

Altschul, 5. F., Cish, W., Miller, W., Myers, E. W. & Lipman, D. J. (1990). Basic local alignment search tool. J MoZ Biol215,403-410.

Baltz, R. H. & Hosted, T. J. (1996). Molecular genetic methods for

Regulation of undecylprodigiosin production improving secondary-metabolite production in actinomycetes. Trends Biotechnol 14,245-250. Bibb, M. J. (1996). The regulation of antibiotic production in Streptomyces coelicolor A3(2). Microbiology 142, 1335-1344. Bibb, M. J., Findlay, P. R. & Johnson, M. W. (1984). The relationship between base composition and codon usage in bacterial genes and its use in the simple and reliable identification of protein coding sequences. Gene 30, 157-166. Brian, P., Riggle, P., Santos, R. A. & Champness, W. C. (1996).

Global negative regulation of Streptomyces coelicolor antibiotic synthesis mediated by an absA-encoded putative signal transduction system. J Bacteriol 178, 3221-3231. Brown, K. L., Wood, 5. & Buttner, M. J. (1992). Isolation and characterisation of the major vegetative RNA polymerase of Streptomyces coelicolor A3(2); renaturation of a sigma subunit using GroEL. Mol Microbiol6, 1133-1139. Bruton, C. J., Guthrie, E. P. & Chater, K. F. (1991). Phage vectors that allow monitoring of transcription of secondary metabolism genes in Streptomyces. Bio/Technology 9, 652-656. Bystrykh, L. V., Fern6ndet-Moren0, M. A., Herrema, J. K., Malpartida, F., Hopwood, D. A. & Dijkhuizen, L. (1996). Production of

actinorhodin-related blue pigments by Streptomyces coelicolor A3(2). J Bacteriol 178, 2238-2244. Chakraburtty, R. & Bibb, M. J. (1997). The ppGpp synthetase gene (relA)of Streptomyces coelicolor A3(2) plays a conditional role in antibiotic production and morphological differentiation. J Bacteriol 179, 5854-5861. Chakraburtty, R., White, J., Takano, E. & Bibb, M. J. (1996).

Cloning, characterization and disruption of a (p)ppGpp synthetase gene (relA) of Streptomyces coelicolor A3(2). Mol Microbiol 19,357-368. Champness, W. C. & Chater, K. F. (1994). Regulation and integration of antibiotic production and morphological differentiation in Streptomyces spp. In Regulation of Bacterial Differentiation, pp. 61-93. Edited by P. Piggot, C. P. Moran & P. Youngman. Washington, DC : American Society for Microbiology. Chater, K. F. (1990). The improving prospects for yield increase by genetic engineering in antibiotic-producing streptomycetes. Bio/ Technology 8, 115-121. Chater, K. F. & Bibb, M. J. (1997). Regulation of bacterial antibiotic production. In Biotechnology, Volume 7, Products o f Secondary Metabolism, pp. 57-105. Edited by H. Kleinkauf & H. von Dohren. Weinheim, Germany: VCH. Chater, K. F., Bruton, C. J., King, A. A. & Suaret, J. E. (1982). The expression of Streptomyces and Escherichia coli drug resistance determinants cloned into the Streptomyces phage &31. Gene 19, 21-32. Coco, E. A., Narva, K. E. & Feitelson, J. 5. (1991). New classes of Streptomyces coelicolor A3(2) mutants blocked in undecylprodigiosin (Red) biosynthesis. Mol Gen Genet 227,28-32. Covarrubias, L. & Bolivar, F. (1982). Construction and characterisation of new cloning vehicles VI. Plasmid pBR329, a new derivative of pBR328 lacking the 482 base pair inverted duplication. Gene 17,79-89. Devereux, J., Haeberli, P. & Smithies, 0. (1984). A comprehensive set of sequence analysis programs for the VAX. Nucleic Acids Res 12,387-395. Dyson, P. & Schrempf, H. (1987). Genetic instability and DNA amplification in Streptomyces lividans 66. J Bacteriol 169, 47964303. Feitelson, J. S,. Malpartida, F. & Hopwood, D. A. (1985). Genetic

and biochemical characterization of the red gene cluster of Streptomyces coelicolor A3(2). J Gen Microbiol 131, 2431-2441. Felsenstein, J. (1989). PHYLIP - phylogeny inference package (version 3.2). Cladistics 5, 164-166. FernBndez-Moreno, M. A., Caballero, J. L., Hopwood, D. A. 81 Malpartida, F. (1991). The act cluster contains regulatory and antibiotic export genes, direct targets for translational control by the bldA transfer RNA gene of Streptornyces. Cell 66, 769-780. Gramajo, H. C., Takano, E. & Bibb, M. J. (1993). Stationary-phase production of the antibiotic actinorhodin in Streptomyces coelicolor A3(2) is transcriptionally regulated. Mol Microbiol 7, 837-845. Guthrie, E. P. & Chater, K. F. (1990). The level of a transcript required for production of a Streptomyces coelicolor antibiotic is conditionally dependent on a transfer RNA gene. J Bacterioll72, 6189-6193. Hackenbeck, R. & Stock, J. B. (1996). Analysis of two-component signal transduction systems involved in transcriptional regulation. Methods Enzymol273,281-300. Hopwood, D. A., Bibb, M. J., Chater, K. F., Kieser, T., Bruton, C. J., Kieser, H. M., Lydiate, D. J., Smith, C. P., Ward, J. M. & Schrempf, H. (1985). Genetic Manipulation of Streptomyces: a

Laboratory Manual. Norwich : John Innes Foundation. Hopwood, D. A., Chater, K. F. & Bibb, M. J. (1995). Genetics of antibiotic production in Streptomyces coelicolor A3(2), a model streptomycete. In Biochemistry and Genetics o f Antibiotic Production, pp. 65-102. Edited by L. C. Vining & C. Stuttard. Newton, MA : Butterworth-Heinemann. Janssen, G. R., Ward, J. M. & Bibb, M. J. (1989). Unusual transcriptional and translational features of the aminoglycoside phosphotransferase gene (aph)from Streptomyces fradiae. Genes Dev 3,415-429. Kieser, T. & Hopwood, D. A. (1991). Genetic manipulation of Streptomyces - integrating vectors and gene replacement. Methods Enzymol204,430-458. Kieser, T. & Melton, R. E. (1988). Plasmid pIJ699, a multi-copy positive-selection vector for Streptomyces. Gene 65, 83-91. Lawlor, E. J., Baylis, H. A. & Chater, K. F. (1987). Pleiotropic morphological and antibiotic deficiencies result from mutations in a gene encoding a tRNA-like product in Streptomyces coelicolor A3(2). Genes Dev 1, 1305-1310. Leskiw, B. K., Bibb, M. J. & Chater, K. F. (1991a). The use of a rare codon specifically during development. Mol Microbiol 5, 28612867. Leskiw, B. K., Lawlor, E. J., FernBndez-Abalos, 1. M. & Chater, K. F. (1991b). TTA codons in some genes prevent their expression

in a class of developmental, antibiotic-negative, Streptomyces mutants. Proc Natl Acad Sci USA 88,2461-2465. Leskiw, B. K., Mah, R.,

Lawlor, E. J. & Chater, K. F. (1993).

Accumulation of bldA-specified transfer RNA is temporally regulated in Streptornyces coelicolor A3(2). J Bacteriol 175, 1995-2005, Madduri, K. & Hutchinson, C. R. (1995). Functional characterization and transcriptional analysis of the dnrR2 locus, which controls daunorubicin biosynthesis in Streptomyces peucetius. J Bacteriol 177, 1208-1215. Malpartida, F., Niemi, J., Navarrete, R. & Hopwood, D. A. (1990).

Cloning and expression in a heterologous host of the complete set of genes for biosynthesis of the Streptomyces coelicolor antibiotic undecylprodigiosin. Gene 93, 91-99. Martlnez-Costa, 0. H., Arias, P., Romero, N. M., Parro, V., Mellado, R. P. & Malpartida, F. (1996). A relA/spoT homologous

737

E. P. GUTHRIE a n d OTHERS

gene from Streptomyces coelicolor A3(2) controls antibiotic biosynthetic genes. J Biol Chem 271, 10627-10634. Merrick, M. J. (1976). A morphological and genetic mapping study of bald colony mutants of Streptomyces coelicolor. J Gen Microbiol96, 299-315. Murray, M. G. (1986). Use of sodium trichloroacetate and mung bean nuclease to increase sensitivity and precision during transcript mapping. Anal Biochem 158, 165-170. Narva, K. E. & Feitelson, J. S. (1990). Nucleotide sequence and transcriptional analysis of the redD locus of Streptomyces coelicolor A3(2). J Bacteriol 172,326-333. Otten, 5. L., Ferguson, J. & Hutchinson, C. R. (1995). Regulation of daunorubicin production in Streptomyces peucetius by the dnrR2 locus. J Bacteriol 177, 1216-1224. Page, R. D. M. (1996). TREEVIEW, an application to display phylogenetic trees on personal computers. CABZOS 12,357-358. Pao, G. M. & Saier, M. H. (1995). Response regulators of bacterial signal transduction systems, selective domain shuffling during evolution. J Mol Evol 40, 136-154. Passantino, R., Puglia, A. M. & Chater, K. F. (1991). Additional copies of the actZZ regulatory gene induce actinorhodin production in pleiotropic bld mutants of Streptomyces coelicolor A3(2). J Gen Microbiol 137, 2059-2064. Piret, J. M. & Chater, K. F. (1985). Phage-mediated cloning of bldA, a region involved in Streptornyces coelicolor morphological development and its analysis by genetic complementation. J Bacteriol 163, 965-972. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989). Molecular Cloning: a Laboratory Manual, 2nd edn. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory. Sanger, F., Nicklen, 5. & Coulson, A. R. (1977). DNA sequencing with chain-terminating inhibitors. Proc Natl Acad Sci USA 74, 5463-5467.

Strohl, W. R. (1992). Compilation and analysis of DNA sequences

associated with apparent streptomycete promoters. Nucleic Acids Res 20,961-974. Stutzman-Engwall, K. J., Otten, 5. & Hutchinson, C. R. (1992).

Regulation of secondary metabolism in Streptomyces spp. and overproduction of daunorubicin in Streptomyces peucetius. J Bacteriol 174, 14+154. Takano, E., Gramajo, H. C., Strauch, E., Andres, N., White, J. & Bibb, M. J. (1992). Transcriptional regulation of the redD

transcriptional activator gene accounts for growth-phase-dependent production of the antibiotic undecylprodigiosin in Streptomyces coelicolor A3(2). Mol Microbiol6, 2797-2804. Thompson, J. D., Higgins, D. G. &Gibson, T. J. (1994). CLUSTAL w,

improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res 22, 4673-4680.

Tsao, S. W., Rudd, B. A. M., He, X., Chang, C. & Floss, H. G. (1985).

Identification of a red pigment from Streptomyces coelicolor A3(2) as a mixture of prodigiosin derivatives. J Antibiot 38, 128-130. Volz, K. (1995). Structural and functional constraints in response regulators. In Two-Component Signal Transduction, pp. 53-64. Edited by J. A. Hoch & T. J. Silhavy. Washington DC : American Society for Microbiology. White, J. & Bibb, M. J. (1997). The bldA-dependence of undecylprodigiosin production in Streptomyces coelicolor A3(2) involves a pathway-specific regulatory cascade. J Bacteriol 179,627-633. Wright, F. & Bibb, M. J. (1992). Codon usage in the G+C-rich Streptomyces genome. Gene 113,5545. Yanisch-Perron, C., Vieira, J. & Messing, J. (1985). Improved M13 phage cloning vectors and host strains : nucleotide sequences of the M13mp18 and pUC19 vectors. Gene 33, 103-119.

Sermonti, G., Petris, A., Micheli, M. R. & Lanfaloni, L. (1977). A

factor involved in chloramphenicol resistance in Streptomyces coelicolor A3(2): its transfer in the absence of the fertility factor. J Gen Microbiol 100,347-353.

738

Received 16 May 1997; revised 20 October 1997; accepted 30 October 1997.